Insights into the critical role of NADPH oxidase(s) in the normal and dysregulated pancreatic beta cell
- First Online:
It is now widely accepted that reactive oxygen species (ROS) contribute to cell and tissue dysfunction and damage in diabetes. The source of ROS in the insulin secreting pancreatic beta cells has traditionally been considered to be the mitochondrial electron transport chain. While this source is undoubtedly important, we fully describe in this article recent information and evidence of NADPH oxidase-dependent generation of ROS in pancreatic beta cells and identify the various isoforms that contribute to O2•− and H2O2 production in various conditions. While glucose-stimulated ROS generation may be important for acute regulation of insulin secretion, at higher levels ROS may disrupt mitochondrial energy metabolism. However, ROS may alter other cellular processes such as signal transduction, ion fluxes and/or cell proliferation/death. The various beta cell isoforms of NADPH oxidase (described in this review) may, via differences in the kinetics and species of ROS generated, positively and negatively regulate insulin secretion and cell survival.
KeywordsDiabetes Insulin secretion Metabolism NADPH oxidase Pancreatic beta cells Reactive oxygen species Review
Glucose-stimulated insulin secretion
Inducible nitric oxide synthase
NOX activator 1
NOX organiser 1
Phosphatidyl inositol 3-kinase
Protein kinase B
Ras-related C3 botulinum toxin substrate
Reactive oxygen species
Regulation of insulin secretion
The consequence of limited scavenging systems is that upon Ca2+ stimulation of mitochondrial and NOX systems, H2O2 concentrations in beta cells may increase rapidly and so easily reach damaging levels. The beta cell must thus strictly limit ROS production to ensure that appropriate levels are maintained, which are essential to and supportive of stimulation of insulin production and release, but not harmful to the cells.
The NADPH oxidase complex
To better explain the role of some specific isoforms in the beta cell, we will now describe the structure and organisation of the NOX family. The NOX family is a group of plasma or sub-cellular membrane-associated superoxide (O2•−)-generating enzymes found in a wide variety of eukaryotic cells. These enzymes are members of a superfamily of flavocytochromes, which transport electrons through biological membranes and catalyse the cytosolic NADPH-dependent reduction of molecular oxygen to O2•− . The NOX family members are related through their common ability to produce O2•− but differ in their membrane and cytosolic protein components.
The phagocyte NOX is a multicomponent enzyme complex, which plays an essential role in host defence. It is composed of two membrane components, three cytosolic proteins and a small GTPase ras-related C3 botulinum toxin substrate (Rac)2 . The membrane components are the catalytic gp91phox (alternatively termed NOX2) and the regulatory p22phox. The catalytic subunit of the respiratory burst oxidase is located in phagocytic cell intracellular phagosome membrane and in the plasma membrane [15, 16]. The membrane component gp91phox has six transmembrane domains and is unstable in the absence of p22phox. The two components gp91phox and p22phox stabilise one another in a tightly associated heterodimer complex, referred to as flavocytochrome b558 . Activation of NOX2 requires translocation of cytosolic factors to the gp91phox/p22phox complex . The cytosolic components are the proteins p40phox, p47phox and p67phox. Translocation of cytosolic regulatory factors to the membrane is required for enzyme activation and O2•− production. When the cell is exposed to stimuli such as various growth factors, cytokines, hormones or bacterial components (such as fMetLeuPhe), the subsequent activation of various isoforms of protein kinase C leads to phosphorylation of the cytosolic component p47phox, which, along with p67phox and p40phox, interacts with the flavocytochrome b558 at the membrane. Concomitantly, Rac dissociates from a RhoGDP dissociation inhibitor and is activated by exchange of GDP for GTP. Rac regulates the activity of the oxidase by a two-step process, directly via contact with gp91phox and by interaction with p67phox [19, 20]. The electron transfer from NADPH to flavin adenine dinucleotide (FAD) and then to molecular oxygen is regulated by the activation domain of p67phox [21, 22]. NOX2 production has been reported in non-phagocytic cells [18, 23].
The homologues of the phagocytic NOX are now known as the NOX family, which comprises seven members: NOX1, NOX2, NOX3, NOX4, NOX5, DUOX1 and DUOX2 . According to structural and functional characteristics, the animal NOX family enzymes can be classified into three groups: (1) the NOX1 to NOX4 group, characterised by the requirement for interaction with p22phox; (2) the NOX5 group, characterised by two calcium-binding EF-hand motifs in the cytosolic N-terminal region; and (3) the DUOX group (DUOX1 and DUOX2), characterised by EF-hand calcium-binding domain and NH2-terminal peroxidase domain [24, 25].
Recently, novel cytoplasmic members of the NOX family enzymes with significant homology to p47phox and p67phox were identified, including an isoform of p47phox, designated NOX organiser 1 (NOXO1) [26, 27], and a homologue of p67phox termed NOX activator 1 (NOXA1) .
p40phox was originally detected by co-immunoprecipitation with p47phox and p67phox, and is a non-glycosylated cytosolic protein with an apparent molecular mass of 40 kDa. p40phox has been shown to interact with p47phox and p67phox with a 1:1:1 stoichiometry .
NOX1 was the first homologue of gp91phox (NOX2) to be described . The small GTPase Rac may be involved in the regulation of NOX1 activity [31, 32]. NOX4 is a gp91phox homologue, which exhibits about 39% homology to NOX2 . A particular feature of NOX4 is the high level of constitutive production of hydrogen peroxide, which is partially dependent on the presence of p22phox [34, 35]. Unlike NOX1, 2 and 3, NOX4 is a constitutively active enzyme and is activated without the need for the GTPase Rac or the cytosolic components , and regulated by protein expression level . Recent studies have demonstrated the participation of NOX4 in the regulation of cell differentiation such as in human and mouse pre-adipocytes [37, 38, 39], thus suggesting a role for H2O2 production in adipocyte differentiation and maturation. While NOX4 expression has been reported in pancreatic islets , the role of NOX4 in beta cells has yet to be determined.
NADPH oxidase isoforms and components in pancreatic islets
Site(s) of ROS generation in the pancreatic beta cell
A recent paper has described the importance of mitochondrial ROS production for GSIS . The authors described experiments in which rat islets were incubated in the presence of glucose and inhibitors of mitochondrial electron transfer (rotenone or antimycin) or a mitochondrial uncoupler, CCCP. In conditions in which mitochondrial function had been disrupted by the latter pharmacological agents, so enhancing ROS production in conditions of reduced or halted ATP generation, insulin secretion was increased. The authors concluded that mitochondrial ROS production is a necessary stimulus for glucose-induced insulin secretion. However, in all conditions tested in the latter paper, NOX-generated ROS would also be present and would also be attenuated by the use of antioxidants. Thus while it is possible to conclude that ROS are essential for glucose-induced insulin secretion, the key site(s) of generation and action are yet to be conclusively determined.
A more widely held view is that mitochondrial dysfunction and ROS overproduction are involved in molecular mechanisms of cell dysfunction in diabetes (see also below) [46, 47]. Mild uncoupling can substantially decrease mitochondrial ROS generation and reduce oxidative stress. The uncoupling mechanism involves activation of uncoupling protein (UCP) 2, which protects cells and enhances GSIS. However, substantial overactivation of UCP 2 disrupts cellular energy metabolism, thus reducing insulin secretion. The critical role of UCP 2 in promotion of pancreatic beta cell insulin secretion was recently illustrated in a study making use of Ucp2−/− mice, in which impaired beta cell function due to enhanced oxidative stress was demonstrated .
Superoxide in solution dismutates to H2O2 with a rate constant of 105 mol (l/s)−1 at pH 7 and this rate can be increased (by approximately four orders of magnitude) due to catalysis by SOD . The biological action of NOX derived-ROS (which can be released extracellularly) may be dependent on the action of SOD and on subsequent diffusion of H2O2 through the plasma membrane or possible transport through selective membrane channels.
Although transmembrane flux of ROS has been proposed, it is tempting to speculate that NOX1/NOX4 or NOX2 could also deliver electrons to intracellular oxygen, so forming intracellular O2•−, which could immediately impact on signalling (Fig. 2). Considering the short half-life of ROS, its participation as a signalling molecule is probably dependent on the site of its generation, which would support the hypothesis of a key role for membrane-associated NOX in signal transduction at the plasma membrane . Regarding the importance of H2O2, previous work has demonstrated a H2O2 dependency of platelet-derived growth factor signalling in smooth muscle cells . In addition, an improvement in insulin signalling was reported to be due to NOX-derived H2O2, which caused protein-tyrosine phosphatase inhibition in adipocytes . Recently, the authors of this review demonstrated that GSIS in rat pancreatic islets was dependent on NOX activity and formation of H2O2 .
Considering the chemical differences between various ROS (e.g. O2•−, H2O2) and indeed reactive nitrogen species, there are likely to be different redox targets for each to transmit its effect. For example NO• binds to the haem group of various enzymes to modulate their activity; O2•− reacts with iron-sulphur clusters of key cellular proteins to alter their activity; and H2O2 can modify proteins primarily through oxidation of thiol groups, which is favoured when these groups are in the deprotonated form. Thiols are important biological targets of oxidation, as they form part of essential antioxidant defences (e.g. glutathione, thioredoxin and glutaredoxin) and are critical for the chemical action of many enzymes, transcription factors and regulatory proteins . While it is difficult to define the relative contribution of sub-cellular structures and associated functions to ROS-associated positive or negative effects on insulin secretion, we would speculate that mitochondrial generation of metabolic stimulus–secretion coupling factors and plasma membrane-associated signal transduction events (including ion-channel activity) are the major sites of interaction (Fig. 3)
Beta cell redox regulation, metabolism and insulin secretion
Chronic exposure to high glucose levels and/or high lipid levels in vivo and in vitro results in glucotoxicity or glucolipotoxicity, whereby beta cell insulin secretion is reduced and the cell is more susceptible to apoptotic stimuli [4, 46, 53]. Under glucotoxic conditions elevated generation of mitochondrial O2•− will occur over an extended period, resulting in overstimulation of UCP 2, mitochondrial proton gradient uncoupling and reduced ATP generation. Transcription factors have been reported to be down-regulated by H2O2, due to decreased expression and/or DNA binding. Duodenal homeobox-1 (PDX-1), a beta cell-specific transcription factor that functions as a potent activator of insulin gene transcription, and MafA are two targets of H2O2 regulation . These transcriptional factors play a crucial role in pancreas development and both beta cell differentiation and function. A recent report described that adenoviral overexpression of dominant negative type c-Jun N-terminal kinase-1 preserved PDX-1 DNA-binding activity in face of oxidative stress and protected insulin gene expression and secretion . Under glucolipotoxic or indeed lipotoxic conditions, elevations in fatty acid oxidation cause extended elevation of mitochondrial generation of O2•− and H2O2 with the associated negative consequences, as described above. In addition generation of diacylglycerol or monoacylglycerol from the elevated intracellular fatty acids associated with the lipotoxic condition may chronically stimulate NOX generation of O2•− and H2O2.
Diabetic C57BL/KsJ-db/db mice were reported to have a significantly greater beta cell mass after antioxidant treatment, possibly due to suppression of apoptosis. They had significantly higher levels of insulin content and insulin mRNA, which was correlated with higher levels of PDX-1 and associated translocation, clearly visible in the nuclei of islet cells after the antioxidant treatment .
With respect to pancreatic beta cell lines, RINm5F cells overexpressing catalase, glutathione peroxidase and Cu/Zn SOD were protected from a lethal pro-inflammatory cytokine mixture (IL-1β, TNF-α and interferon-γ) and exhibited a significantly lower level of ROS-dependent protein damage . Indeed earlier reports, using primary rat islets, described the importance of redox enzymes to the regulation of beta cell metabolism, including a study describing the inhibition of glutathione reductase (a GSSG-reducing enzyme), which impaired glucose metabolism and insulin secretion . GSIS was reported to be dependent on the level of intracellular reduced glutathione . Using a patch-clamp technique, investigators reported a reduction in ATP-sensitive K+ channel activity, suppression of plasma membrane electrical excitability and inhibition of metabolism after H2O2 addition [60, 61, 62].
More recent studies have identified that mitochondrial aconitase, mitochondrial adenine-nucleotide translocase  and cytosolic glyceraldehyde-3-phosphate dehydrogenase (GAPDH) are susceptible to oxidative modification [64, 65]. One of four cysteine side chains of GAPDH is especially important to the catalytic reaction process (Cys 149) and is located near to the NAD coenzyme in the active site of the enzyme . Mitochondrial aconitase belongs to a family of iron/sulphur-containing dehydratases, whose activities depend on the redox state of the [4Fe–4S]2+ cluster. It has been shown that exposure of purified mitochondria to oxidants, particularly O2•− and H2O2, resulted in enzyme inactivation . Inactivation appeared to result from interactions between aconitase and a mitochondrial membrane component responsive to H2O2 . GAPDH inhibition may divert metabolites to the hexosamine pathway  or possibly to polyol and advanced glycation end-product formation . The possible mechanisms by which high levels of ROS inhibit beta cell metabolism and insulin secretion have been summarised in Fig. 1. Under specific conditions, high rates of glucose metabolism prevented the accumulation of superoxide-derived ROS in primary beta cells .
ROS-associated beta cell damage in diabetes
Insulin resistance and pancreatic beta cell insufficiency are major features in the progression of type 2 diabetes mellitus [71, 72]. The molecular basis for excessive mitochondrial oxidative damage in diabetes has been expertly reviewed elsewhere .
Interestingly, compared with many other cell types, the beta cell may be at high risk of oxidative damage with an increased sensitivity for apoptosis. This high risk may be due to: (1) excessive levels of mitochondrial O2•− generation; (2) additional O2•− and H2O2 generation through elevated beta cell NOX activity (as described here); and (3) limited antioxidant defences. With respect to type 2 diabetes, beta cell dysfunction and associated depressed insulin secretion must be evident before hyperglycaemia develops . It is clear that excessive levels of glucose, lipid, endocrine and various inflammatory factors interact at the level of the pancreatic islet to promote beta cell dysfunction. Thus any explanation of molecular integration at the level of the beta cell must include common mechanisms of dysfunction. One possible candidate mechanism is the activation of NOX and consequent ROS production. The beta cell NOX is activated by glucose, saturated fatty acids (such as palmitic acid), endocrine factors (angiotensin II) and some proinflammatory cytokines . The evidence for cytokine involvement in type 2 diabetes mellitus is still under investigation, but it is widely accepted that cytokines play a major role in beta cell dysfunction and destruction in type 1 diabetes mellitus.
Impairment of beta cell function in vivo may involve excessive generation of O2•− and H2O2 through increased NOX activity as discussed here and elsewhere . This would subsequently affect mitochondrial function, reducing ATP production and insulin secretion [11, 69]. Additionally palmitate metabolism may give rise to ceramide synthesis, ceramide being a key component of the signal transduction pathway for ROS-induced apoptosis . Ceramide may also induce apoptosis by inactivating pro-survival pathways. Thus this lipid can inhibit phosphatidyl inositol 3-kinase (PI3K), which in turn results in blockade of protein kinase B (PKB, also known as Akt/PKB) activation . Downstream targets of the PI3K/PKB pathway involved in survival include glycogen synthase kinase-3 (GSK-3) , caspase-9  and the Bcl-2 family member Bad . The mechanism for the effect of high glucose on pancreatic beta cell apoptosis remains to be elucidated, but several studies have been conducted on this subject. Thus exposure of islets from diabetes-prone Psammomys obesus (obese sand rat, an animal model of type 2 diabetes) to high glucose levels resulted in a dose-dependent increase in beta cell DNA fragmentation suggesting caspase activation . Addition of high glucose resulted in apoptosis of pancreatic beta cells from ob/ob mice and normal Wistar rats via a Ca2+-dependent process in vitro .
The proinflammatory cytokine IL1-β is known to induce apoptosis of islet beta cells (associated with type 1 diabetes) by stimulating cytochrome c release from the mitochondria with subsequent activation of downstream caspases. Proinflammatory cytokine mixtures (especially IL-1β, TNF-α and IFN-γ) are typically required to induce death in rodent cell lines. The molecular mechanisms of cytokine-induced cell death appear to involve upregulation of the enzyme inducible nitric oxide synthase (iNOS) and thus nitric oxide generation and also NOX subunit expression resulting in ROS production . O2•− and NO can combine to generate peroxynitrite (ONOO−) , which can induce mitochondrial dysfunction, endoplasmic reticulum stress and apoptosis . Oxidative stress due to elevated O2•− and H2O2 and subsequent peroxynitrite generation may damage DNA by promoting single-strand DNA break formation, leading to initiation of the apoptotic cascade. Indeed, evidence has recently been obtained from the laboratory of one of the authors of this review that pharmacological inhibition of NOX or iNOS will attenuate cytokine-induced inhibition of insulin secretion from mouse islets (M. Michalska, G. Wolf, R. Walther and P. Newsholme, unpublished data) suggesting a potential role for peroxynitrite formation in inhibition of insulin secretion.
Experimental overexpression of free radical dissipating enzymes such as catalase, glutathione peroxidase and the cytoplasmic copper–zinc SOD protects cultured rodent beta cells from deleterious combinations of cytokines (IFN-γ, TNF-α and IL-1β) . The alloxan-induced diabetes-susceptible (ALR/Lt) mouse, originally selected for resistance to alloxan (a beta cell-selective toxin that generates hydroxyl radicals ), provides a rare example of systemically upregulated antioxidant defences that include beta cells. The molecular basis for this resistance was a systemic upregulation of molecules and enzymes associated with free radical dissipation [84, 85]. The lower oxidative burden in ALR/Lt mice correlated with significantly higher specific activities for a battery of antioxidant enzymes not only in comparison with control mice, but also in comparison with the closely related NOD/Lt strain, which develops spontaneous T cell-mediated type 1 diabetes [85, 86]. ALR pancreatic beta cells, both in vivo and in vitro, were resistant to alloxan and also resistant in vitro to the cytotoxic combination of IFN-γ, TNF-α and IL-1β. However, a recent review has highlighted that the mechanisms of alloxan-induced cellular cytotoxicity are complex and therefore results obtained from the ALR/Lt mouse model must be evaluated in the context of the complexity of action of alloxan .
Evidence has been published that impaired insulin and insulin like growth factor signalling pathways contribute to the cellular mechanisms underlying the pathology of Alzheimer’s disease . Recent work has shown that ROS generation in the brain of patients with Alzheimer’s disease is primarily via amyloid-β-activated microglial NOX. Melatonin provided significant protective effects due to inhibition of phosphorylation of the p47phox subunit of NOX via a PI3K/PKB-dependent signalling pathway, also blocking the translocation of p47phox and p67phox subunit to the membrane, as well as downregulating the binding of p47phox to gp91phox, thereby impairing the assembly of the functional NOX enzyme . Conflicting reports exist on the effect of melatonin on insulin secretion, but it is tempting to speculate that it may alter NOX assembly and activity as described for microglial cells. This is of importance as genetic studies have demonstrated a link between variations in MTNR1B gene, hyperglycaemia, impaired early-phase insulin secretion and beta cell function .
While ROS appear to be beneficial for the redox-dependent regulation of insulin secretion, chronic production and elevation of ROS above a critical threshold level may lead to activation of oxidative stress concomitantly with reduced insulin secretion. As NOX is expressed at relatively high levels in the islet beta cell, relative to other islet cells, potential therapies based on targeting NOX in the islet may be beneficial for maintaining beta cell function and integrity in conditions of high glucose, lipid or inflammatory challenge, and therefore could be effective in type 2 diabetes treatment.
This work was funded by The Health Research Board of Ireland and Enterprise Ireland, as well as by The State of São Paulo Research Foundation (FAPESP), the Programme for support for centres of excellence PRONEX and the National Council for Scientific and Technological Development (CNPq), Brazil.
Duality of interest
The authors declare that there is no duality of interest associated with this manuscript.